• The taxon-independent 1:1 ortholog classes are superhelpful; we use these more than anything. The PRO already does a great job at creating these. However, there appear to be some organisms many of whose orthologs aren’t properly classified within these taxon-independent orthologs. For example, ‘calcineurin subunit B type 1 (rat)’ (P63100) is a subclass only of ‘Rattus norvegicus protein’ (000029033), while it appears it should (also) be a subclass of the taxon-independent ‘calcineurin subunit B type 1’ (000013150). I’m guessing the PRO folks know about this already and just haven’t gotten around to it yet.
• What would be really helpful are many more classes more general than the 1:1 ortholog classes, as families/abstractions are also frequently mentioned in the literature. The PRO already has some useful family classes we’ve used, e.g., ‘AKT kinase’ (000029189), ‘aurora kinase’ (000035365), ‘death receptor’ (000021999), ‘G-protein coupled receptor’ (000030035), ‘smad protein’ (000000027), ‘collagen alpha chain’ (000003262), and ‘mitogen-activated protein kinase’ (000000019) (though it looks like a lot of the MAPKs aren’t subsumed by this?). Attached is a list of abstraction classes we’ve created for CRAFT; these are currently defined as unions of proper PRO classes, but it’d be totally fine if these were just superclasses of the corresponding existing PRO classes we’ve listed. (For most of the PRO classes, the shorter gene acronyms are used for convenience.) Not all of these may be appropriate for the PRO as proper families; we’ll defer to the discretion of the PRO curators. These examples were specifically created for annotation within CRAFT, but a much wider catalog of such families/abstractions would be really useful for larger-scale mining of the literature.
